The goal of this project is to improve the quality of cause of death information for public health research and planning. The study will provide a more scientific basis for using cause of death data needed for estimating global and regional mortality patterns. A related broad goal is to study the transition in non-communicable disease mortality rates to shed light on key controversy in public health: whether age-specific death rates for non-communicable diseases rise or decline with economic development. The first component of the project will focus on developing analytical procedures to correct for miscertification of deaths form ischemic heart disease, cancers and other major chronic diseases. Multiple regression methods will independent variables having a predictable relationship to disease levels (e.g. cigarette consumption) will be applied to national cause of death statistics to develop disease-specific correction algorithms to apply to registered cause of death data. In order to make better use of cause of death data collected in sentinel surveillance systems by verbal autopsies of relatives of the deceased, validation of cause of death data from sample sites in China and Tanzania will be carried out. Validation methods will include growth-balance demographic techniques to estimate completeness of death recording by age and sex. Verification of the reliability of cause of death information will be carried out using repeat verbal autopsies of family members (China) and validation against hospital clinical diagnosis (Tanzania). A critical component of the project will be to use corrected cause of death data to analyze long-tern trends in age-specific non-communicable disease mortality rates to examine the extent to which income-elastic risk factors for major non-communicable diseases affect mortality trends. We will firstly describe mortality trends back to 1900 for selected industrialized and middle income countries using national sources. We will then apply the correction algorithms developed in the project to re- estimate the apparent trend in age-specific NCD mortality rates over the last 100 years in these countries to test the research hypothesis.